De Groote et al, 2013 – “Measures of health sciences journal use…”
Access to paid journal subscriptions is a major cost – and subject of study – for libraries of every professional nature, though particularly the health sciences. Sorting high-demand and low-demand, high-importance sources from those which cost much but provide little is a crucial task in library management, and one made more difficult by disparate data.
Metrics of use for journal access come primarily from three sources: local citations, vendor reports, and link-resolver statistics for online activity. These three together must be analyzed to produce a useful picture of the importance and cost of each of a given institution’s subscriptions. Given the complexity of such models, however, it is vital to understand the relation between each of their three constituents. The ability to analyze the whole of the model from a limited input set or to identify discrepancies in expected values, as any data specialist would attest, is quite valuable.
De Groote et al. suggest that, in most cases, link-resolver data may be swapped in for vendor data, greatly simplifying the process and providing reasonable accuracy (with a Spearman rank order correlation of 84.3%).
The authors acknowledge that this model does result in some number of anomalies; however, they posit that the presence and nature of such anomalies in a given circumstance may suggest to a competent librarian the existence of issues with their institution’s vendors or online resources. Further analysis of this unexplained variance, they suggest, may also yield insights into user behavior and guide additional studies.
A cursory read of their work in this article would be wise for any individual responsible for library management decisions pertaining to online resources and subscription set optimization.